虚拟脊柱侧凸病例腰椎自动生成方法

doi:10.19678/j.issn.1000-3428.0067788

摘要/Abstract

摘要： 统计形状模型(SSM)是一种描述可变形物体形态学变化的模型,被广泛应用于三维(3D)肝脏分割、3D脊柱生成等任务。针对现有SSM在生成虚拟青少年特发性脊柱侧凸(AIS)腰椎病例数据时存在生成结果不真实且对AIS病例适用性较差的问题,提出一种基于三维变分自编码生成式对抗注意力网络(3D-VGAN)的虚拟AIS腰椎自动生成方法。3D-VGAN模型由编码器、生成器和判别器组成,编码器和生成器在变分自编码器(VAE)模型的基础上结合空间注意力机制提取数据特征,同时利用残差模块解决了神经网络训练过程中的网络退化问题。为了克服生成-判别结构的3D-VGAN模型中存在的稳定性不强、生成器与判别器训练速度不匹配的问题,使用阈值训练法对3D-VGAN模型进行训练,提升了稳定性。在基于43例AIS病例腰椎3D模型的数据集上的实验结果表明:3D-VGAN模型在病例重建实验中的结构相似性(SSIM)系数为0.999 62,相比3D-VAE、3D-变分自编码生成式对抗注意力网络(VAEGAN)和SSM模型分别提高了0.080 09%、0.002 00%、0.122 20%;在病例生成实验中的弗雷歇初始距离(FID)系数为275.653 48,相比3D-VAE、3D-VAEGAN和SSM模型分别降低了8.420 55%、0.977 73%、7.319 27%。上述实验结果验证了3D-VGAN模型通过注意力机制、残差模块和阈值训练法取得了更好的虚拟AIS病例腰椎数据生成性能。

关键词: 青少年特发性脊柱侧凸, 生成式对抗网络, 统计形状模型, 图像生成, 深度学习

Abstract: Statistical Shape Model (SSM) describe the morphological variations in deformable objects and are widely used in Three-Dimensional (3D) liver segmentation, 3D spine generation, and other fields. However, the existing SSMs yield inaccurate results and demonstrate unsatisfactory applicability when generating virtual Adolescent Idiopathic Scoliosis (AIS) lumbar spine cases. An automatic generation method of lumbar spine using virtual AIS based on a 3D Variational autoencoder Generation Adversarial Network (3D-VGAN) is proposed in this study. The 3D-VGAN model comprises an encoder, a generator, and a discriminator. The encoder and generator extract data characteristics based on a Variational AutoEncoder (VAE) model and combine the spatial attention mechanism. Meanwhile, residual block is used to solve network degradation during neural network training. To overcome the insufficient stability of the network model and the mismatch in training speed between the generator and discriminator, a threshold training method is used to train the network model to improve its stability. Experimental results on a dataset based on a 3D lumbar spine model of 43 AIS cases show that the Structural Similarity (SSIM) coefficient of this model is 0.999 62, which is 0.080 09%, 0.002 00 %, and 0.122 20% higher than those of 3D-VAE, 3D-Variational AutoEncoder Generation Adversarial Network (VAEGAN), and SSM models, respectively. The Frechet Inception Distance (FID) coefficient of the case generation experiment is 275.653 48, which is 8.420 55%, 0.977 73%, and 7.319 27% lower than those of 3D-VAE, 3D-VAEGAN, and SSM models, respectively. Thus, the proposed model outperforms existing models in lumbar spine data generation using virtual AIS case by leveraging attention mechanism, residual block, and threshold training methods.

Key words: Adolescent Idiopathic Scoliosis(AIS), Generation Adversarial Network(GAN), Statistical Shape Model(SSM), image generation, deep learning

中图分类号:

TP39

赵逸飞, 张俊华. 虚拟脊柱侧凸病例腰椎自动生成方法[J]. 计算机工程, 2024, 50(5): 323-329.

ZHAO Yifei, ZHANG Junhua. Automatic Generation Method of Lumbar Spine Using Virtual Scoliosis Cases[J]. Computer Engineering, 2024, 50(5): 323-329.

https://www.ecice06.com/CN/Y2024/V50/I5/323

参考文献

[1] NEGRINI S, DONZELLI S, AULISA A G, et al. 2016 SOSORT guidelines:orthopaedic and rehabilitation treatment of idiopathic scoliosis during growth[J]. Scoliosis and Spinal Disorders, 2018, 13:3.
[2] 海涌, 王云生. 青少年特发性脊柱侧凸:探索、发展、创新[J]. 骨科, 2023, 14(1):1-3. HAI Y, WANG Y S. Adolescent idiopathic scoliosis:exploration, development and innovation[J]. Orthopaedics, 2023, 14(1):1-3.(in Chinese)
[3] BREDBENNER T L, ELIASON T D, FRANCIS W L, et al. Development and validation of a statistical shape modeling-based finite element model of the cervical spine under low-level multiple direction loading conditions[J]. Frontiers in Bioengineering and Biotechnology, 2014, 2:58.
[4] MELLER S, KALENDER W A. Building a statistical shape model of the pelvis[J]. International Congress Series, 2004, 1268:561-566.
[5] BRYAN R, SURYA MOHAN P, HOPKINS A, et al. Statistical modelling of the whole human femur incorporating geometric and material properties[J]. Medical Engineering & Physics, 2010, 32(1):57-65.
[6] NICOLELLA D P, BREDBENNER T L. Development of a parametric finite element model of the proximal femur using statistical shape and density modelling[J]. Computer Methods in Biomechanics and Biomedical Engineering, 2012, 15(2):101-110.
[7] SARKALKAN N, WEINANS H, ZADPOOR A A. Statistical shape and appearance models of bones[J]. Bone, 2014, 60:129-140.
[8] WAGNER D, KAMER L, ROMMENS P M, et al. 3D statistical modeling techniques to investigate the anatomy of the sacrum, its bone mass distribution, and the trans-sacral corridors[J]. Journal of Orthopaedic Research, 2014, 32(11):1543-1548.
[9] ARMSTRONG J R, CAMPBELL J Q, PETRELLA A J. A comparison of Cartesian-only vs. Cartesian-spherical hybrid coordinates for statistical shape modeling in the lumbar spine[J]. Computer Methods and Programs in Biomedicine, 2021, 204:106056.
[10] HEIMANN T, MEINZER H P. Statistical shape models for 3D medical image segmentation:a review[J]. Medical Image Analysis, 2009, 13(4):543-563.
[11] RASOULIAN A, ROHLING R, ABOLMAESUMI P. Lumbar spine segmentation using a statistical multi-vertebrae anatomical shape+pose model[J]. IEEE Transactions on Medical Imaging, 2013, 32(10):1890-1900.
[12] SOZOU P D, COOTES T F, TAYLOR C J, et al. A non-linear generalisation of PDMs using polynomial regression[C]//Proceedings of Conference on British Machine Vision. New York, USA:ACM Press, 1994:397-406.
[13] SOZOU P D, COOTES T F, TAYLOR C J, et al. Non-linear point distribution modelling using a multi-layer perceptron[C]//Proceedings of 1995 British Conference on Machine Vision. New York, USA:ACM Press, 1995:107-116.
[14] WILMS M, HANDELS H, EHRHARDT J. Multi-resolution multi-object statistical shape models based on the locality assumption[J]. Medical Image Analysis, 2017, 38:17-29.
[15] 侯佳英. 基于深度学习的图像生成算法研究与实现[D]. 北京:北京邮电大学, 2021. HOU J Y. Research and implementation on image synthesis algorithm based on deep learning[D]. Beijing:Beijing University of Posts and Telecommunications, 2021. (in Chinese)
[16] HO J, JAIN A, ABBEEL P. Denoising diffusion probabilistic models[EB/OL].[2023-05-14]. https://doi.org/10.48550/arXiv.2006.11239.
[17] 王燕萍, 吕磊, 苏志龙, 等. 基于深度学习的高质量图像生成方法综述[J]. 激光杂志, 2023, 44(6):7-12. WANG Y P, LV L, SU Z L, et al. Overview of high-quality image generation methods based on deep learning[J]. Laser Journal, 2023, 44(6):7-12.(in Chinese)
[18] KINGMA D P, WELLING M. Auto-encoding variational Bayes[EB/OL].[2023-05-14]. https://arxiv.org/abs/1312.6114.
[19] 郭奉琦, 孟凡荣, 王志晓. 基于变分自编码器的谣言立场分类算法[J]. 计算机工程, 2022, 48(2):99-105. GUO F Q, MENG F R, WANG Z X. Rumor stance classification algorithm based on variational auto-encoder[J]. Computer Engineering, 2022, 48(2):99-105.(in Chinese)
[20] GOODFELLOW I, POUGET-ABADIE J, MIRZA M, et al. Generative adversarial nets[EB/OL].[2023-05-14]. https://arxiv.org/abs/1406.2661.
[21] 罗嗣卿, 陈慧. 基于生成对抗网络的图像场景转换[J]. 计算机工程, 2023, 49(4):217-225. LUO S Q, CHEN H. Image-scene transformation based on generative adversarial networks[J]. Computer Engineering, 2023, 49(4):217-225.(in Chinese)
[22] 陈佛计, 朱枫, 吴清潇, 等. 生成对抗网络及其在图像生成中的应用研究综述[J]. 计算机学报, 2021, 44(2):347-369. CHEN F J, ZHU F, WU Q X, et al. A survey about image generation with generative adversarial nets[J]. Chinese Journal of Computers, 2021, 44(2):347-369.(in Chinese)
[23] LARSEN A, SNDERBY S K, LAROCHELLE H, et al. Autoencoding beyond pixels using a learned similarity metric[C]//Proceedings of International Conference on Machine Learning. New York, USA:ACM Press, 2016:1558-1566.
[24] VASWANI A, SHAZEER N, PARMAR N, et al. Attention is all you need[C]//Proceedings of the 31st International Conference on Neural Information Processing Systems. New York, USA:ACM Press, 2017:6000-6010.
[25] HE K M, ZHANG X Y, REN S Q, et al. Deep residual learning for image recognition[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition(CVPR). Washington D.C., USA:IEEE Press, 2016:770-778.
[26] HEUSEL M, RAMSAUER H, UNTERTHINER T, et al. GANs trained by a two time-scale update rule converge to a local Nash equilibrium[EB/OL].[2023-05-14]. https://arxiv.org/abs/1706.08500.
[27] BARRATT S T, SHARMA R. A note on the inception score[EB/OL].[2023-05-14]. https://arxiv.org/abs/1801.01973.
[28] WANG Z, BOVIK A C, SHEIKH H R, et al. Image quality assessment:from error visibility to structural similarity[J]. IEEE Transactions on Image Processing, 2004, 13(4):600-612.
[29] ANGELO L D, STEFANO P D. A new method for the automatic identification of the dimensional features of vertebrae[J]. Computer Methods and Programs in Biomedicine, 2015, 121(1):36-48.

选择文件类型/文献管理软件名称

选择包含的内容